Cognitive reflection is a distinct and measurable trait

Abstract

We report the findings of an adversarial collaboration examining whether the cognitive reflection test (CRT) measures anything beyond mathematical aptitude and, if so, whether its incremental predictive validity can be attributed to reflection, per se. We found that an 8-item CRT has greater predictive validity than an 8-item Mathematical Aptitude Test (MAT) consisting of comparably difficult items which lack dominant intuitive lures. Further, the incremental predictive validity stems from the CRT's measurement of reflection, which we show using both structural equation models and a dual-response paradigm that helps distinguish susceptibility to intuitions from inadequate mathematical aptitude.